San Diego-On the third day of AACR 2024, most presenters had thrown away their posters as the exhibit floor was drawing to a close. But in one of the dozens of rows, several conversations were centered around a single poster. Several members of Genialis are using machine learning to model fundamental aspects of his KRAS cancer biology and develop RNA-based biomarkers to predict tumor response and clinical benefit to KRAS inhibitors. was explaining his krasID.
My luck was with Mr. Aditya Pai, Head of Business Development, and Mr. Sami Thakriti, Business Development Manager. They described to me a poster that led to results demonstrating accurate prediction of clinical benefit in real-world non-small cell lung cancer (NSCLC). ) Patient cohort treated with sotorasib (Lumaclas). But while Mr. Pai and his Mr. Takriti's guidance was helpful, it wasn't necessary. Because the data is very clear. krasID predicts clinical response to the KRAS G12C inhibitor sotorasib in a real-world non-small cell lung cancer patient cohort.
“Responsive person” or “Non-responsive person”
Genialis launched in 2017 with a $2.3 million seed round to advance visual exploration of next-generation sequencing data. This led to the development of the Genialis Expressions software. This enables machine learning-driven biomarker discovery by aggregating consistently analyzed and annotated data.
But over the years, Genialis has transformed into a computational precision medicine company dedicated to unraveling complex biology and finding new ways to combat disease. Genialis is developing next-generation patient classifiers that use machine learning and high-throughput omics data to understand underlying disease biology and predict how patients will respond to targeted therapies. I started focusing on
The first iteration of this effort was ResponderID. This machine learning platform was designed to identify new biomarkers for diagnostic tests as well as drug discovery and discovery programs. ResponderID includes predictions of how effective treatments are in peritumoral regions, classification of microsatellite instability (MSI), tumor mutational burden (TMB), and other cutting-edge immune signatures, and KRAS inhibition. It is equipped with algorithms that can support the development of new drugs such as drugs. . Behind ResponderID, Genialis raised over $13 million in his 2023 Series A.
Within the ResponderID framework, Genialis developed krasID based on a computational tool called a “classifier.” These machine learning algorithms automatically order or classify data into a set of one or more “classes.” One of the most common examples is an email classifier that scans emails and filters them by the class label “spam” or “not spam.” Classifiers have also appeared in pop culture and entertainment, most notably on the show “Silicon Valley” as a tool to classify images as “hot dogs” or “non-hot dogs.”
Personalize KRAS inhibitor treatment
The krasID classifier was designed to classify patient RNA-seq data into “responders” or “non-responders.” In reality, krasID does not provide a binary answer, but instead generates a probability of response to KRAS inhibition.
To do this, Genialis used RNA sequencing to measure gene expression in tumor tissue, including FFPE samples, and discovered a number of biological “modules” related to or close to KRAS. For example, one biological module represented a cell-dependent, gene expression signature that distinguishes tumors that depend on KRAS for survival and pathway activation. The krasID classifier was trained using five of these models to predict the probability of response to KRAS inhibition.
After training, krasID was used on gene expression data from a group of real-world NSCLC patients before treatment with sotorasib (Lumacras) to infer how patients would respond to treatment. Grouping patients by krasID score (“krasID-high” or “krasID-low”) splits the average Kaplan-Meier survival curve into two survival curves, one shifted toward increasing survival and one shifted toward increasing survival. one shifts in the direction of decreasing survival. Specifically, the median survival time of patients classified as “krasID-high” (338 days) was more than twice that of patients classified as “krasID-low” (158 days). did.
